**2.6. Sample preparation for chemical analysis**

82 New Approaches to the Study of Marine Mammals

**2.4. Stable Isotopes Analysis (SIA)** 

**2.5. Trophic level estimations** 

carbon flux and sediments [40].

using the following algorithm [37, 38]:

*CONSUMER*

*TP*

(‰) were determined using the following equation [21,26]:

Vienna), for *δ*15*N*. The equivalent equation for *δ*13*C* isotope ratios (‰) is:

isotopically homogeneous than samples, was ± 0.19‰ for *δ*13*C* and ±0.24 for *δ*15*N*.

Carbon and nitrogen isotopic analyses on fish biopsies and Galapagos sea lion hair were accomplished by continuous flow, isotopic ratio mass spectrometry (CF-IRMS) using a GV-Instruments® IsoPrime attached to a peripheral, temperature-controlled, EuroVector® elemental analyzer (EA) (University of Winnipeg Isotope Laboratory, UWIL). One-mg samples were loaded into tin capsules and placed in the EA auto-sampler along with internally calibrated carbon/nitrogen standards. Nitrogen and carbon isotope results are expressed using standard delta (*δ*) notation in units of per mil (‰).The delta values of carbon (*δ*13*C*) and nitrogen (*δ*15*N*) represent deviations from a standard. *δ*15*N* isotope ratios

*δ*15*N* = [(15*N*/14*NSAMPLE*/15*N*/14*NSTANDARD*) – 1] x 1000 where 15*N*/14*N*SAMPLE is the isotope ratio of the tissue sample analyzed; and, 15*N*/14*N*STANDARD represents the ratio of the international standard of atmospheric *N*2 (air), IAEA-N-1 (IAEA,

*δ*13*C* = [(13*C*/12*CSAMPLE*/13*C*/12*CSTANDARD*) – 1] x 1000 The standard used for carbon isotopic analyses was the Vienna PeeDee Belemnite (VPDB). Analytical precision, determined from the analysis of duplicate samples, was ±0.13‰ for *δ*13*C* and ±0.6‰ for *δ*15*N*. The analytical precision based on standards, which are more

The trophic positions (TPCONSUMER) of the prey species (i.e. fish) and the predator (Galapagos sea lion) were determined relative to the baseline *δ*15*N* (assumed to occupy a trophic level 2),

> <sup>15</sup> <sup>15</sup> δ - δ = + 2 3.4 *CONSUMER BASELINE*

*N N*

Where *δ*15*N*CONSUMER is the average *δ*15*N* signature value of the predator; *δ*15*N*BASELINE is the *δ*15*N* signature at the base of the food web; and 3.4‰ is the isotopic, trophic level enrichment factor (*∆*15*N*), recommended to be used for constructing food webs when a priori knowledge of *∆*15*N* is unavailable [39]. The *δ*15*N*BASELINE was established as the *δ*15*N* signature of the particulate organic matter (POM) of bottom sediments in the eastern equatorial Pacific Ocean (250 km south of the islands) with a value of 5.5‰ [31, 40], which is relatively close to the *δ*15*N* value of 7.3‰, reported recently for phytoplankton in the Galapagos [30]. The rationale for using this signature is supported by the fact that the assimilation of nitrogen (i.e., NO3¯) up taken from near surface marine waters by phytoplankton is reflected by *δ*15*N* values of POM, which is also a major component of the Contaminant analyses were conducted in the Regional Dioxin Laboratory (RDL) at the Institute of Ocean Sciences (IOS), Fisheries and Ocean Canada (DFO), based on analytical methodologies described elsewhere [44]. In brief, the muscle-blubber biopsy samples of Galapagos sea lion pups (0.053 to 0.212 g wet weight) and subsamples of fish homogenate (9.23 to 10.5 g) were spiked with a mixture of surrogate internal standards which contained all fifteen 13C12-labeled PCBs, and a mixture of labelled organochlorine pesticides (OCPs): D3 1,2,4-Trichlorobenzene, 13C6 1,2,3,4 Tetrachlorobenzene, 13C6 Hexachlorobenzene, 13C6*�*-HCH, 13C6*�*-HCH, 13C10 trans Nonachlor, 13C12 TeCB-47, 13C12*p*,*p*'-DDE, 13C12 Dieldrin, 13C12*o*,*p*-DDD, 13C12*p*,*p*'-DDD, 13C12*o*,*p*-DDT, 13C12*p*,*p*'-DDT, 13C10 Mirex. All surrogate internal standards were purchased from Cambridge Isotope Laboratories (Andover, MA). The spiked samples were homogenized with Na2SO4 in a mortar, transferred quantitatively into an extraction column, and extracted with DCM/hexane (1:1 v/v). For some of the samples the extract formed two layers/phases, a waxy-precipitate layer and the solvent layer. The solvent layer was transferred to a clean flask and the waxy precipitate was treated with several aliquots of hexane and DCM. Each of these was transferred to the flask that contained the solvent layer of the extract. Despite the treatment with additional volumes of hexane and DCM, vortexing and pulverization, the waxy precipitate (for sea lions) did not dissolved in the solvents used and as a result it was not included in the corresponding sample extract that was used for lipid and contaminants determinations.The DCM:Hexane sample extracts were evaporated to dryness and the residue was weighted in order to determine the total lipid in the samples. Subsequently the residue was re-suspended in 1:1 DCM/Hexane and divided quantitatively into two aliquots. The larger aliquot (75% of the extract) was subjected to sample-cleanup for PCBs determinations. The remaining (25% of the extract) was used for OCP determinations.

## **2.7. PCB and OC pesticides analyses**

Sample extracts were analyzed for PCB congeners and target OCPs by gas chromatography/high-resolution mass spectrometry (GC/HRMS). The specific methodology and protocols for the quantification and analytical methods to determine PCB congeners and OCPs have previously been reported in prior published papers (34, 35).

### **2.8. Quality assurance/quality control measures**

The mass spectrometry conditions used for all the analyses, the composition of the linearity calibration solutions, the criteria used for congener identification and quantification and the quality assurance – quality control procedures used for the quantification of PCBs and OCPs followed those described in detail elsewhere [34, 35, 44].

Assessing Biomagnification and Trophic Transport of Persistent Organic Pollutants in the Food Chain of the Galapagos Sea Lion (*Zalophus wollebaeki*): Conservation and Management Implications 85

> log *PREDATOR PREY PREDATOR PREY -*

*([C ]/[C ])*

10

Where *Cpredator* and *Cpre*y are appropriately normalized (e.g., lipid normalized) chemical concentrations in the predator and prey, and TL*predator* and TL*prey* are the trophic levels of the predator and prey. In essence, the BMFTL is the biomagnification factor normalized to a single trophic level increase in the food-web [45].The use of trophic magnification factors (TMFs) is currently an emerging approach to better assess the biomagnification of POPs in marine food webs [16]. An important number of studies in the northern hemisphere have relied on the use of the TMF for this purpose [15, 16, 18]. Thus, the use of TMF coupled with stable isotope analysis (SIA) to track the amplification and transport of POPs in food webs is a recommended methodology in eco-toxicology to study the biomagnification of POPs. The lack of prey samples and minimal trophic levels required (≥ 3) precluded to undertaking a trophic magnification factor (TMF) assessment in this

Concentrations of all detected POPs were blank corrected using the method detection limit (MDL), defined as the mean response of the levels measured in three procedural blanks used plus three times the standard deviation (SD) of the blanks (MDL = Meanblanks + 3\*SDblanks). Following this methodology, the concentration of each PCB congener and OC pesticide was determined based on concentrations above the MDL only. Only PCBs detected in 100% of samples and above the MDL were used for data analysis and calculations of BMFs. Contaminant concentration data were log-transformed to fit the assumption of normality criterion before statistical analysis. ∑PCB concentrations were calculated as the sum of PCB-52, PCB 74, PCB 95, PCB-99, PCB-101, PCB-105, PCB-118, PCB 128, PCB - 138/163/164**,** PCB-146, PCB 153, PCB 156, PCB 174, PCB 180, PCB 183, PCB 187, PCB 201 and PCB 202. ∑DDTs were dened as the sum of *o, p'*-DDE, *p, p'*-DDE, *o, p'*-DDD, *p, p'*-DDD, *o, p'*-DDT and *p, p'*-DDT, and ∑chlordanes as the sum of *trans*-chlordane, *cis*-chlordane, *trans*-

To further support the analysis of biomagnification of POPs in the tropical food chain of the Galapagos, statistical comparisons between the concentrations of selected PCBs (e.g., PCBs 153, 180), ∑DDTs, *p*,*p*'-DDE and other organochlorine pesticides measured in the Galapagos sea lion and those detected in diet items (i.e., mullet and thread herring) were conducted. These comparisons were conducted using analyses of variance (ANOVA) if variances were homoscedastic (i.e., equal variances) or Welch's analyses of variance if variances or standard deviations were heteroscedastic (i.e., unequal variances as tested by Levene's test or Bartlett test, *p*< 0.05), and a Tukey-Kramer honestly significant difference (HSD) test, which is a post-hoc method recommended to test differences between pairs of means among groups that contain unequal sample sizes [46]. Inter-site comparisons among rookeries samples

10

*TL TL BMF*

*TL \* =*

**2.10. Data treatment and supporting statistical analysis** 

study.

nonachlor and *cis*-nonachlor.

### **2.9. Bioaccumulation parameter**

In general, the biomagnification of contaminants is basically quantified as the biomagnification factor in terms of the concentration of a given chemical in the consumer or predator relative to the concentration in the diet or prey (i.e. BMF= *C*B/*C*D, where *C*B is the chemical concentration in the organism and *C*D is the chemical concentration of the diet). To quantify biomagnification in the Galapagos sea lions relative to prey items (i.e., thread herring and mullet) and to explore the effect of the magnitude of trophic level differences on the BMF measures, the predator-prey biomagnification factor (BMF TL) was used for data interpretation in this study.The criterion applied to indicate the capability of the chemical to biomagnify was a BMF > 1. A BMF statistically greater than 1 indicates that the chemical is a probable bioaccumulative substance [7].

## *2.9.1. Predator-prey Biomagnification Factor (BMF TL)*

Following this approach, the mean lipid normalized concentration of each contaminant measured in Galapagos sea lion pups was divided by the mean lipid adjusted concentration in the prey. Then, the biomagnification factor can be adjusted to represent exactly one trophic level in difference using the trophic level estimated from *δ*15*N*. Therefore, the field based predator-prey biomagnification factor normalized to trophic position or BMFTROPHIC LEVEL (BMF*TL*) is calculated using the following equation [15]:

$$\mathcal{BMF}\_{\text{TL}} = \frac{(\mathcal{C}\_{\text{normmacro}}/\mathcal{C}\_{\text{max}})}{T L\_{\text{normnorm}} - T L\_{\text{norm}}}$$

Where *Cpredator* and *Cpre*y are chemical concentrations in the predator and prey, expressed in units of mass of chemical (*μ*g) per kg of the predator and mass chemical (*μ*g) per kg of prey in a lipid normalized basis (i.e. BMFLIPID WEIGHT),and TL predator and TLprey are the trophic levels of the predator and prey. The BMF*TL* values were used to measure biomagnification in the tropical food chain between two adjacent trophic levels (i.e., the difference in TL between predator and prey is small), assuming steady state in contaminant concentrations between predator and prey. Since BMF*TL* can be related to the trophic magnification factor (TMF), which describes the increase of contaminants from one trophic level to the other (derived from the slope, b, of the relationship between an organism's log lipid normalized chemical concentration), it can also be expressed as BMF*TL\** as proposed by Conder *et al*. [45]:

Assessing Biomagnification and Trophic Transport of Persistent Organic Pollutants in the Food Chain of the Galapagos Sea Lion (*Zalophus wollebaeki*): Conservation and Management Implications 85

$$\mathbf{BMF}\_{\mathrm{TL}} \, \mathbf{\hat{}} = \mathbf{10} \left[ \frac{\log\_{10}{\left( \mathbf{C}\_{\mathrm{rxn}, \mathrm{rxn}} \right) / \left( \mathbf{C}\_{\mathrm{rxn}} \right)} \right]$$

Where *Cpredator* and *Cpre*y are appropriately normalized (e.g., lipid normalized) chemical concentrations in the predator and prey, and TL*predator* and TL*prey* are the trophic levels of the predator and prey. In essence, the BMFTL is the biomagnification factor normalized to a single trophic level increase in the food-web [45].The use of trophic magnification factors (TMFs) is currently an emerging approach to better assess the biomagnification of POPs in marine food webs [16]. An important number of studies in the northern hemisphere have relied on the use of the TMF for this purpose [15, 16, 18]. Thus, the use of TMF coupled with stable isotope analysis (SIA) to track the amplification and transport of POPs in food webs is a recommended methodology in eco-toxicology to study the biomagnification of POPs. The lack of prey samples and minimal trophic levels required (≥ 3) precluded to undertaking a trophic magnification factor (TMF) assessment in this study.

#### **2.10. Data treatment and supporting statistical analysis**

84 New Approaches to the Study of Marine Mammals

**2.9. Bioaccumulation parameter** 

probable bioaccumulative substance [7].

BMF*TL\** as proposed by Conder *et al*. [45]:

*2.9.1. Predator-prey Biomagnification Factor (BMF TL)* 

LEVEL (BMF*TL*) is calculated using the following equation [15]:

**2.8. Quality assurance/quality control measures** 

followed those described in detail elsewhere [34, 35, 44].

The mass spectrometry conditions used for all the analyses, the composition of the linearity calibration solutions, the criteria used for congener identification and quantification and the quality assurance – quality control procedures used for the quantification of PCBs and OCPs

In general, the biomagnification of contaminants is basically quantified as the biomagnification factor in terms of the concentration of a given chemical in the consumer or predator relative to the concentration in the diet or prey (i.e. BMF= *C*B/*C*D, where *C*B is the chemical concentration in the organism and *C*D is the chemical concentration of the diet). To quantify biomagnification in the Galapagos sea lions relative to prey items (i.e., thread herring and mullet) and to explore the effect of the magnitude of trophic level differences on the BMF measures, the predator-prey biomagnification factor (BMF TL) was used for data interpretation in this study.The criterion applied to indicate the capability of the chemical to biomagnify was a BMF > 1. A BMF statistically greater than 1 indicates that the chemical is a

Following this approach, the mean lipid normalized concentration of each contaminant measured in Galapagos sea lion pups was divided by the mean lipid adjusted concentration in the prey. Then, the biomagnification factor can be adjusted to represent exactly one trophic level in difference using the trophic level estimated from *δ*15*N*. Therefore, the field based predator-prey biomagnification factor normalized to trophic position or BMFTROPHIC

> *TL <sup>=</sup> - (C / C ) BMF TL TL*

Where *Cpredator* and *Cpre*y are chemical concentrations in the predator and prey, expressed in units of mass of chemical (*μ*g) per kg of the predator and mass chemical (*μ*g) per kg of prey in a lipid normalized basis (i.e. BMFLIPID WEIGHT),and TL predator and TLprey are the trophic levels of the predator and prey. The BMF*TL* values were used to measure biomagnification in the tropical food chain between two adjacent trophic levels (i.e., the difference in TL between predator and prey is small), assuming steady state in contaminant concentrations between predator and prey. Since BMF*TL* can be related to the trophic magnification factor (TMF), which describes the increase of contaminants from one trophic level to the other (derived from the slope, b, of the relationship between an organism's log lipid normalized chemical concentration), it can also be expressed as

*PREDATOR PREY PREDATOR PREY* Concentrations of all detected POPs were blank corrected using the method detection limit (MDL), defined as the mean response of the levels measured in three procedural blanks used plus three times the standard deviation (SD) of the blanks (MDL = Meanblanks + 3\*SDblanks). Following this methodology, the concentration of each PCB congener and OC pesticide was determined based on concentrations above the MDL only. Only PCBs detected in 100% of samples and above the MDL were used for data analysis and calculations of BMFs. Contaminant concentration data were log-transformed to fit the assumption of normality criterion before statistical analysis. ∑PCB concentrations were calculated as the sum of PCB-52, PCB 74, PCB 95, PCB-99, PCB-101, PCB-105, PCB-118, PCB 128, PCB - 138/163/164**,** PCB-146, PCB 153, PCB 156, PCB 174, PCB 180, PCB 183, PCB 187, PCB 201 and PCB 202. ∑DDTs were dened as the sum of *o, p'*-DDE, *p, p'*-DDE, *o, p'*-DDD, *p, p'*-DDD, *o, p'*-DDT and *p, p'*-DDT, and ∑chlordanes as the sum of *trans*-chlordane, *cis*-chlordane, *trans*nonachlor and *cis*-nonachlor.

To further support the analysis of biomagnification of POPs in the tropical food chain of the Galapagos, statistical comparisons between the concentrations of selected PCBs (e.g., PCBs 153, 180), ∑DDTs, *p*,*p*'-DDE and other organochlorine pesticides measured in the Galapagos sea lion and those detected in diet items (i.e., mullet and thread herring) were conducted. These comparisons were conducted using analyses of variance (ANOVA) if variances were homoscedastic (i.e., equal variances) or Welch's analyses of variance if variances or standard deviations were heteroscedastic (i.e., unequal variances as tested by Levene's test or Bartlett test, *p*< 0.05), and a Tukey-Kramer honestly significant difference (HSD) test, which is a post-hoc method recommended to test differences between pairs of means among groups that contain unequal sample sizes [46]. Inter-site comparisons among rookeries samples followed the same statistical methods. Statistical comparison tests were conducted at a level of significance of *p*< 0.05 (α = 0.05).

Assessing Biomagnification and Trophic Transport of Persistent Organic Pollutants in the Food Chain of the Galapagos Sea Lion (*Zalophus wollebaeki*): Conservation and Management Implications 87

Observed concentrations of selected POPs in Galapagos sea lion and two of its main prey items are summarized in Table 1. Galapagos sea lions represented the largest number of organisms sampled in this study (*n* = 41) and exhibited the highest concentrations of PCBs and OC pesticides. The multi-comparison post hoc analysis, including sea lions and prey fish, showed that no significant differences in OC pesticides and PCB congener concentrations were observed between male and female pups. Fish prey commonly exhibited significantly lower concentrations than Galapagos sea lion pups (ANOVA and

Concentrations of ∑DDTs in Galapagos sea lions ranged from 16.0 to 1700 *μ*g/kg lipid and ∑DDTs were the predominant OC pesticide in Galapagos sea lion pups, as previously reported [35]. ∑Chlordanes were the second most abundant group of contaminants present. *Trans*-nonachlor represented 68% of ∑chlordanes, followed by *cis*-chlordane, *cis*nonachlor and *trans*-chlordane (Table 1), a pattern comparable to that reported in pups of southern elephant seals (*Mirounga leonina*) [49] and Weddell seals (*Leptonychotes weddellii*) [50]. This indicates that *trans*-nonachlor is a predominant chlordane compound in

Within the hexachlorocyclohexanes (HCHs), *β*-HCH was the only isomer detectable in all pups (>MDL). *β*-HCH was the dominant HCH isomer in blubber samples of California sea lions (*Zalophus californianus*) from Baja California [51] and in toothed cetaceans from tropical and temperate waters of the Indian and North Pacific oceans [52] due to the greater biomagnification of the most bioaccumulative *β*-HCH versus *γ*-HCH [3, 20]. Interestingly, the mean *β*-HCH concentration in Galapagos sea lions was higher than the mean ∑HCH concentrations measured in spinner dolphins (*Stenella longirostris*) (21.3 *μ*g/kg lipid) captured in a marine area of the Eastern Tropical Pacific [52] in offshore waters north of the

Both dieldrin and mirex were detected in all pups with concentrations ranging from 0.85 to 24 *μ*g/kg lipid for mirex and from 9.00 to 83.0 *μ*g/kg lipid for dieldrin. Concentrations of ∑PCBs (i.e., sum of 20 PCB congeners) ranged between 16.0 and 380 (*μ*g/kg lipid) in pups

OC pesticides, including ∑DDTs, chlordanes, *β*-HCH, dieldrin and mirex, and individual PCB congeners detected in Galapagos sea lion pups were also detected (> MDL) in all sampled thread herring and mullet prey samples. Significantly lower concentrations of OC pesticides and PCBs were found in thread herrings and mullets than in Galapagos sea lion pups (ANOVA and multi-comparisons Tukey-Kramer (HSD) post-hoc test, *p*< 0.05; Table 1). PCB 202 was the only congener exhibiting similar concentrations in sea lions and

and from 1.0 to 140 (*μ*g/kg lipid) in fish preys (Table 1).

multi-comparisons Tukey-Kramer (HSD) post-hoc test, *p*< 0.05) (Table 1, Figure 2).

**3.2. POP concentrations in animals and inter-site comparisons** 

*3.2.1. Galapagos sea lions* 

pinnipeds.

Galapagos.

*3.2.2. Fish prey* 

Principal Components Analyses (PCAs) were conducted on the fractions of PCBs and organochlorine pesticides relative to total concentrations by contaminant group (i.e., contaminants expressed as a fraction of total) for each sample to visualize spatial differences in patterns in sea lion pups from different sites within the Galapagos Archipelago and elucidate potential sources (i.e., local versus global-atmospheric). First, samples with undetectable values were replaced by a random number between the lowest and the highest concentration that were detectable (> MDL) to account for uncertainty before PCA (i.e., *trans*-chlordane and PCB 110 showed zero values in blanks in three and two samples out of 20, respectively; therefore; there was not possible to calculate MDLs), or otherwise removed from the PCAs. Secondly, samples were normalized to the concentration total before PCA to remove artefacts related to concentrations differences between samples. Finally, the centered log ratio transformation (division by the geometric mean of the concentration-normalized sample followed by log transformation) was then applied to this compositional data set to produce a data set that was unaffected by negative bias or closure [47]. Regressions, statistical comparisons and PCAs were run using JMP 7.0 (SAS Institute Inc.; Cary, NC, USA).
